Evidence suggests that humans might have neurological specializations for music processing, but a compelling adaptationist account of music and dance is lacking. The sexual selection hypothesis cannot easily account for the widespread performance of music and dance in groups (especially synchronized performances), and the social bonding hypothesis has severe theoretical difficulties. Humans are unique among the primates in their ability to form cooperative alliances between groups in the absence of consanguineal ties. We propose that this unique form of social organization is predicated on music and dance. Music and dance may have evolved as a coalition signaling system that could, among other things, credibly communicate coalition quality, thus permitting meaningful cooperative relationships between groups. This capability may have evolved from coordinated territorial defense signals that are common in many social species, including chimpanzees. We present a study in which manipulation of music synchrony significantly altered subjects' perceptions of music quality, and in which subjects' perceptions of music quality were correlated with their perceptions of coalition quality, supporting our hypothesis. Our hypothesis also has implications for the evolution of psychological mechanisms underlying cultural production in other domains such as food preparation, clothing and body decoration, storytelling and ritual, and tools and other artifacts.